3D Processing

Usage of 3D Information

In the past, most robotic applications concentrated on the processing of 2D information about the environment, such as laser data or visual images. In the last years however, the usage of 3D sensors has increased signigicantly. At this point in time, almost every rescue robot uses such a sensor. 3D information provides much more possibilities for cognitive systems. A basic advantage is the computation of three-dimensional positions in space, which can for example be used for the localization of objects and provide the robot with information about its own positioning within the world.

Perception of Environment

The GETbot is equipped with an ASUS Xtion Pro Live depth sensor. This sensor combines a visual camera and a depth camera and creates colored three-dimensional point clouds, which describe the structure of the environment. Point clouds are published with up to 60 Hertz and with a maximum resolution of 640x480 points.

Applications

Up to now the three-dimensional point clouds are used to analyze the structure of the underground around the robot and to classify its practicability. The GETbot can only drive on even terrain and flat ramps. Larger steps, gaps and irregularities of the ground are not accessible for the robot and therefore are marked impassable. The resulting information is used by the navigation system to avoid accidents and damage to the robot.

Further Goals

Currently further applications using 3D data are in progress. A main goal is to achieve a 3D reconstruction of the whole environment using colored point clouds. Furthermore the grasping of objects with a robot arm will be simplified by a 3D object recognition system. Therefore, the position, orientation and size of graspable objects are computed.